ARTICLE

Ultra high damage threshold optics for highpower lasersYurina Michine1 & Hitoki Yoneda1*

The output energies of lasers have increased year-by-year since their invention. Compared to

this increase of laser energies, the damage threshold of optical components has not strongly

changed. Therefore, the size of optics in high-energy laser system increases. This situation

could change dramatically if optics with higher damage threshold were developed. Here, we

propose a high damage threshold optics using a neutral gas as an active medium. More than

95% diffraction efficiency has been achieved. The damage threshold for a 6 ns laser pulse is

measured to be 1.6 kJ/cm2. The aperture size of the present system is about 60 mm2. Based

on this result, we anticipate that control of a 1 kJ laser beam may be achievable using 1 cm

sized optics, driven by a < 50 mJ ultraviolet laser, making this scheme promising in high

power laser applications.

https://doi.org/10.1038/s42005-020-0286-6 OPEN

1 Institute for Laser Science, University of Electro-Communications, 1-5-1 Chofugaoka, Chofu, Tokyo 182-8585, Japan. *email: yoneda@ils.uec.ac.jp

COMMUNICATIONS PHYSICS | (2020) 3:24 | https://doi.org/10.1038/s42005-020-0286-6 |www.nature.com/commsphys 1

1234

5678

90():,;

S ince the conception of lasers, a great deal of effort has beeninvested in order to increase laser power and energy. High-energy lasers are opening up new research fields. These

include new fusion energy sources1, new acceleration methods forcharged particle beams2, and intense extreme ultraviolet (EUV)sources for semiconductor manufacturing3. The output energy oflaser pulses has increased from 100 mJ to 2MJ4 during the last 50years. Despite such rapid growth of the output energy, thedamage threshold of optical elements has not increased at asimilar rate. The damage threshold of dielectric coated mirrorsdepends on the pulse duration and laser wavelength. For thevisible and near-infrared region, typical values are 1–200 J/cm2

for a nanosecond pulse laser5. The damage threshold for struc-tured optics such as a grating is lower than for flat mirrors andtypical damage thresholds are 0.1–12 J/cm2 (refs. 6,7). These smallvalues are the main reason why high-energy lasers need meter-size optics.

Optical damage is a serious problem not only for energetic giantlaser systems but also for high repetition rate lasers. Even wellbelow the nominal optical damage threshold, the damage-risk isnot zero. Recently, laser average power has increased significantlyand laser pulse repetition rates have grown from 10Hz to 10 kHzand more. Due to the probabilistic aspect of the optical damage,the actual useful fluence in the high repetition laser is muchsmaller than the singe-pulse damage limit. Once optical damageoccurs in conventional solid optics, the damaged spot does notrecover and it is not easy to replace the optics in a large-size opticslaser system. Recently, a method for optical damage managementidea has been employed, but the usable intensity is not dramati-cally increased. A typical reduction factor for practical use is from2 to 10 (ref. 8).

This situation can be improved if transient optics are introduced.One candidate is plasma transient optics. To date, plasma gratings9–13

and plasma mirrors14 have been proposed and some are alreadyinstalled in real laser systems. In general, plasma optics use thepolarizability of free electrons to modulate the refractive index.Normally, these electrons are generated with nonlinear opticalinteractions. When a high-density plasma is created in the interactionof an intense pulse laser with a gas or solid surface, the obtainable

refractive index change is around 10−1–10−3. This number is similarto the conventional optical coating. However, the energy needed toproduce free electrons is typically several or tens of electronvolts peratom. This large energy deposition causes a high-energy densitycondition in the plasma optics and results in a short lifetime. The laserpower required to produce the index modulation or change in theoptics is quite large and sometimes even larger than the power of thediffracted laser beams.

To solve these problems, we propose an alternative approach totransient diffraction optics using neutral gas with a linearabsorption process. The damage threshold of a gas medium fornanosecond lasers is normally >1 kJ/cm2 (ref. 15), which is twoorders of magnitude higher than the operating laser energy flux ofconventional solid-state optics. To achieve gas-phase diffractionoptics, a spatial density modulation of the gas is used. However,the typical refractive index of neutral gas is only 10−4 or 10−5

above unity. This low value is one of the main barriers to creatinga real optical device with neutral gas medium. Some ideas toincrease the optical path length were previously proposed16.Instead, we describe here a method to create large density mod-ulations in a neutral gas. In this scheme, the modulation ampli-tude is almost 100% so that we achieve more than a half-wavelength optical path difference with 1 cm thickness of themedium. In addition, due to this large optical depth, the devicecan be used as a volume refractive grating to obtain asymmetricdiffraction. In this case, we can concentrate the diffracted beamentirely in the 1st order diffraction with almost zero intensity for0th and −1st order beams17. In this paper, we propose thetransient grating made by the ozone-mixed neutral gas. It haslarger 1.6 kJ/cm2 damage threshold for 6 ns lasers and enoughhigh (96%) diffraction efficiency.

ResultsPeriodical large density modulation in neutral gas. A schematicdrawing of our system is shown in Fig. 1. We use a mixed gas ofozone (O3) and oxygen (O2) at one atmosphere pressure. Theozone molecule has quite large absorption for deep ultraviolet(UV) light but almost no absorption in the visible and infrared

Fig. 1 Schematic drawing of ozone-mixed gas diffraction optics. The ozone is generated with radiation from the corona discharge and atmosphericdielectric barrier discharge. The ozone-mixed gas enters into the rectangular cross-sectional flow tube. In this tube, the mixed gas flow is almost laminar.At 15 cm downstream, we put the open window which has 6 × 10mm rectangular shape. After this section, the mixed gas is pumped by the diaphram pumpwith constant flow velocity.

ARTICLE COMMUNICATIONS PHYSICS | https://doi.org/10.1038/s42005-020-0286-6

2 COMMUNICATIONS PHYSICS | (2020) 3:24 | https://doi.org/10.1038/s42005-020-0286-6 | www.nature.com/commsphys

region18,19. Therefore, a spatially modulated UV laser (in thefollowing, called the UV writing beam) is used to generate anozone density modulation. When an ozone molecule absorbs aUV photon, it dissociates into atomic oxygen (O) and an oxygenmolecule (O2). At atmospheric pressure, the dissociated oxygenatom almost immediately recombines with another oxygenmolecule and regenerates the ozone20. With this process, theinitial absorbed photon energy is converted into thermal energyof O2 and O3 molecules. If we use a periodic intensity-modulatedUV beam, a periodic temperature modulation is created in O3

and O2 gas with no initial density modulation. This conditionleads to coupled sound and entropy waves in the gas. These wavesthen create the density modulation.

Dynamics of the density modulation. Figure 2 shows the tem-poral history of the ozone density profile after periodic energydeposition with the UV writing beam. In this experiment, a 100-mJ,10-ns pulse from a 248-nm Kr*F excimer laser is used. The ozonedensity profile is measured by absorption measurement with 5-ns,287-nm probe laser, and modulation of the refractive index ismeasured with 5 ns, and 598 nm probe. Even though the UV laser isabsorbed in the ozone gas, there is no density change immediatelyafter the UV pulse. However, the density modulation of the ozonemolecules appears within 50–60 ns. The ozone density increaseoccurs at the destructive interference of the UV writing beam whilethe density decreases for constructive interference. After reachingthe peak modulation, the modulation decreases and increases again.This temporal variation is a simple oscillation and the oscillationperiod is a function of grating fringe spacing, which is determinedby the UV writing beam. The hydrodynamic motion is accuratelydescribed by hydrodynamic calculations. At the time of maximumdensity modulation, the induced variation of the optical path lengthis of the order of 1 μm. This is enough to achieve an efficientdiffractive optics.

Based on the transient refractive index modulation, we havedeveloped a diffraction optics using this medium. The diffractionefficiency is determined by the initial ozone density, themodulating Kr*F excimer laser intensity, thickness of medium,wavelength, and incident angle of the diffracted beam, and by thetime interval between the UV writing beam and the diffracted

beam pulse. Uniformity is also important. In this device, we usesaturable absorption to produce uniform absorption along theincident UV laser. In addition, the optimum matching conditionbetween the period of modulation, thickness, and difference ofrefractive index is adjusted to minimize diffraction of undesiredorders and to achieve higher diffraction efficiency. Afteroptimizing the above conditions, we could succeed to get morethan 95% diffraction efficiency with this transient gas mediumdevice.

Diffraction efficiency. Figure 3 shows measured diffraction effi-ciency of this grating as a function of the UV writing beamintensity in the optimum diffraction condition. In this measure-ment, a 532-nm, 6-ns, 1-mJ pulse from a commercial Q-switchedNd:YAG laser is used for the diffracted laser beam. This laser hassingle mode in space and multi-mode in time. The typical spec-trum width is about λ/Δλ= 104. The density-modulated volumein this experiment is 6 mm × 10mm × 10mm. The diffractionefficiency is defined by the ratio of the 1st order diffraction energyto incident pulse energy. As shown in Fig. 3, the average dif-fraction efficiency is 96% at 63 mJ/cm2 of the UV writing beam.We also check the fluctuation in diffraction efficiency shot by shotand single-shot standard deviation is 4.2%. The temporal windowfor this highest efficiency condition is about 10 ns in this gratingperiod. We succeeded to get stable and high diffraction efficiencywith low-intensity modulating laser beam energy. A two-dimensional diffraction efficiency map is shown in the Methodssection. The wave front of the diffracted beams is kept within λ/10distortion (see Methods section).

Damage threshold. Next, we measure the damage threshold ofthis gas medium grating under the best diffraction condition. Innanosecond laser pulse case, the damage threshold is determinedby the ionization threshold of the gas. The possible nonlineareffects in the gas are discussed in the Methods section. Theoccurrence of ionization damage is observed by emissionfrom the damage point, by a change of intensity profile of thetransmitted laser beam, and by detection of wave front distortion.We use 532 nm and 6 ns laser pulse to check three kinds ofgases (air, and ozone-mixed oxygen gas with and without the UV

Fig. 2 Temporal history of ozone density. a After illumination by the ultraviolet (UV) writing beam. The writing beam has a spatial intensity modulationproduced with interferometry optics. The ozone density at the constructive interference (red line) decreases while the density at the destructiveinterference (blue line) increases. The error bars are determined by maximum and minimum values in tens fringes inside one interferometer image. Thetiming of the maximum modulation depends on the period of the grating. The minimum ozone density at the bright fringe area is almost zero, therefore, alarge density modulation is created. b Two-dimentional modulation images of ozone density at different delay times from the pulse of the UV writing beam.A clear grating structure is seen in the ozone-mixed gas.

COMMUNICATIONS PHYSICS | https://doi.org/10.1038/s42005-020-0286-6 ARTICLE

COMMUNICATIONS PHYSICS | (2020) 3:24 | https://doi.org/10.1038/s42005-020-0286-6 |www.nature.com/commsphys 3

laser). Figure 4 shows the percentage of damage as a function ofinput laser energy. The breakdown threshold of the gratingmedium is 1.6 kJ/cm2. The ozone-mixed gas has slightly lowerdamage threshold due to easy ionization from the excited-stateatoms and molecules. Even though this damage threshold is morethan two orders of magnitude higher than conventional optics. Inaddition, this optical element is transit optics so that the safetyfactor for conventional solid optical elements is no moreconsidered.

DiscussionWe describe a diffraction grating in ozone–oxygen mixed gasprepared by using a UV writing beam. The diffraction efficiencyis as high as 96%, the damage threshold is 1.6 kJ/cm2, and thediffraction control beam is 63 mJ/cm2. The suitability of thisapproach for nanosecond duration pulses has been demonstrated.For picosecond and femtosecond lasers, the damage threshold ofthis grating decreases due to the increased influence of nonlineareffects of the gas, but it is still expected to have several timeshigher damage threshold than the conventional solid grating.Extrapolating from our results acquired with a 6 × 10 mm2

aperture and a 40-mJ Kr*F excimer laser in the linear regime, weanticipate this device may be able to control pulse energies up to1 kJ. The fundamental function of this diffraction grating is highefficient diffraction of a high-energy flux laser beam so that manyoptical elements can be replaced with this grating such as opticalswitching, spatial mode filtering. This could lead to a largereduction in the size of high-energy laser systems.

MethodsOzone production and flow system. For our diffraction device, it is necessary toprepare high-density ozone-mixed gas. We use pure O2 gas (99.5%) for the ozonegeneration and medium of the grating, because the damage threshold of the gasmedium depends on the contained micro and nano particles and the amount ofaerosol15,21–23. In addition, the mixed gas must flow in a laminar way to achieve auniform density of ozone. In our system, an atmospheric pressure discharge systemis used with precooled oxygen gas (typical temperature is 10 °C). In the dischargearea, a radio frequency (13MHz) electric field (~10 kV/mm) is applied. The ozone-mixed gas flows into a flow tube whose cross section is 4–10 mm(along the laserpropagation) × 10 mm(height). After 15 cm of gas-flow, at the middle of the flowtube length, we make 6 mm × 10mm rectangular holes in two facing walls. Thelasers can pass through this windowless section. From the end of the flow tube, wepump out the flow gas with a constant flow rate. The flow rate of the source O2 gasand the pumping speed of the ozone-mixed gas are adjusted to keep constant ozonedensity at the open window area. The typical flow speed is about 5 L/min with10 mm × 10 mm cross section. The ozone density profile at the open window areais checked by absorption of a 287-nm probe laser. The change of the ozone densityin the open window area is kept to be within a few percent as shown in Fig. 5. Theozone density is determined by the flow rate of the gas, temperature of the sourcegas, and discharge power. We vary the ozone fraction from 1 to 10% in thisexperiment to achieve the maximum diffraction efficiency. The main limitation toincrease this aperture size is maintaining laminar gas flow with the windowlesscondition. That means the size of this optics can be increased in the directionperpendicular to the flow axes.

Diffracting laser beam. The ozone molecule has a large absorption cross sectionfor deep UV photons. We use an excimer laser system (Kr*F excimer laser:wavelength of 248 nm, pulse of 15 ns, Coherent COMPEX 101) as a deep UV lightsource. Typical repetitive rate is from 5 to 20 Hz. The periodical intensity profile isprepared with Michelson interferometer. To make a clean high-visibility fringepattern, the excimer laser is operated with narrow band operation with injectionseeding, and the band width of the modulation writing beam is 2–3 cm−1. Theoptics in the interferometer has high surface flatness. (Nominal quality is λ/20 atλ= 633 nm.) When we increase the line density, high temporal coherence of UVlasers and large angles with two UV laser beams are required to achieve sufficientlyhigh contrast spatial modulation. In addition, the duration of the first energydeposition should be shorter than grating period divided by sound velocity, becausethe rise time of the density modulation is order of grating spacing divided by thesound velocity of the medium gas.

The excimer laser has relatively small saturation energy fluence, so the intensityprofile at the laser oscillator exit is quite uniform. In our system, this uniformintensity profile is image-relayed onto the diffraction optics of ozone-mixed gaswith a spatial filter optical system. The output energy of this laser system can bevaried from 50 mJ to 200 mJ with change of operating voltage of Kr*F excimerlaser. Due to the large cross section of the ozone molecule at 248 nm, theabsorption is saturated. Therefore, energy deposition along the diffracting laserbeam in the ozone-mixed gas is almost constant in this experiment. This uniformabsorption in the entire volume of the ozone-mixed gas is a key point to achievestable and high efficient diffraction optics.

Two-dimensional map of diffraction efficiency. Both the UV diffracting laserand the diffracted 0.53 μm laser are synchronized with time accuracy of a fewnanoseconds in this research system. The diffracted beam is from a non-seededQ-switched YAG laser so the actual temporal waveform of this diffracted beam hascomplex spike-like structure and relatively broad band spectrum (λ/Δλ ~104). Even

Fig. 4 The test of the damage threshold of the gas medium with 6 ns and532 nm pulse laser. The damage tests are performed with pure air, ozone-mixed gas, and ozone-mixed gas with deep ultraviolet (UV) laserillumination. The intensity of UV laser is beyond the saturation level of theabsorption. Among these three media, we decide that the damagethreshold of our ozone-mixed gas device is 1.6 kJ/cm2. The error bar in thisfigure means statistical error in the measurement.

Fig. 3 Typical diffraction efficiency as a function of the energy density ofthe ultraviolet (UV) writing beam. Measured diffraction efficiency withaverage value over the vertical axis at the center of the diffraction beam.After 63 mJ/cm2 illumination, the diffraction efficiency reaches as high as96%. The error bars are determined by the standard deviation of200 shots; 4.2% fluctuation automatically decreases when the averageefficiency is close to 100%. There is a trade-off between the fluctuation ofthe diffraction efficiency and the production efficiency for the grating(diffracted beam energy divided by UV writing beam energy).

ARTICLE COMMUNICATIONS PHYSICS | https://doi.org/10.1038/s42005-020-0286-6

4 COMMUNICATIONS PHYSICS | (2020) 3:24 | https://doi.org/10.1038/s42005-020-0286-6 | www.nature.com/commsphys

with these non-ideal laser conditions, 96% maximum diffraction efficiency isachieved. The formation speed of the density modulation in this ozone-mixed gasgrating is similar to the sound velocity of the room temperature gas. Therefore, theduration of the maximum density modulation is long enough compared to thepulse duration of the laser beams and timing accuracy in this system.

A two-dimensional map of the measured 1st order diffraction efficiency isshown in Fig. 6. In this experiment, the cross section of the diffracting beam is 4mm × 10 mm, while the diameter of the diffracted laser beam is 10 mm. The area ofthe diffracted beam is indicated with dotted line in Fig. 6. The center of the UVdiffracting beam is almost on the edge of this circle. As shown in this map, thediffraction efficiency at the center of the diffracting UV laser reaches almost 100%,while this number gradually decreases at the edge of the diffracting beam. This isbecause there is low fringe visibility of two interferometric diffracting UV laser

beams with imperfect temporal coherences. Therefore, this edge reduction of thediffraction efficiency could be improved with higher temporal coherence of thediffracting UV lasers.

Angle dependence of the diffraction efficiency. Measured angle dependence onthe diffraction efficiency is shown in Fig. 7. The condition of the UV laser and thediffraction laser are the same as those of Fig. 3 experiment. The fringe spacing isabout 40 μm. The 80% diffraction efficiency angle tolerance is about 0.06°. Thecalculated diffraction efficiency is also shown in this figure. In this calculation, step-like density modulation is assumed and the integrated phases along the tilted rayare calculated. They are in good agreement. According to the simulation, the

Fig. 6 Diffraction efficiency map. a Two-dimensional map of the measured 1st order diffraction efficiency. The yellow ellipse represents the irradiated areaof the diffracted beam; the horizontal axis profile (b) and the vertical axis profile (c) of diffraction efficiency are also shown on the right.

Fig. 5 Spatial profiles of the ozone density at the open window area. a Transmitted image of probe laser for ozone at the open window area. b Measuredozone density profile at z= 1 mm and z= 7mm inside the open window area. If the flow rates at the upstream and the pumping rate at the downstream areoptimized, the density profile is an almost constant profile in the space.

COMMUNICATIONS PHYSICS | https://doi.org/10.1038/s42005-020-0286-6 ARTICLE

COMMUNICATIONS PHYSICS | (2020) 3:24 | https://doi.org/10.1038/s42005-020-0286-6 |www.nature.com/commsphys 5

typical bandwidth to keep high diffraction condition is as large as 1.5%. It corre-sponds to a 100-fs pulse-width laser.

Wave front of diffraction beam. The wave front of the 0th and 1st order dif-fraction beams is measured by self-referenced interferometry, in which measuredbeam is divided into two branches and one beam, 5× expanded, is interfered withthe original one. To realize an almost perfect single spatial mode condition for thediffraction laser, output beam from a single mode fiber is used. The measuredfringe is analyzed with Zernike polynomials expansion method to determine wave

front distortion quantitatively. The other parameters in this experiment are thesame as those in the maximum diffraction efficiency experiments.

Figure 8 shows the color counter map of the measured wave front for the 0thand the 1st order diffracted beam. It is seen that the wave front aberration of the 1storder diffraction beam is less than λ/10. Because the aberration of the 0th orderbeam is similar to this behavior, we conclude that there was negligible distortion ofthe diffracted beam in our ozone-mixed gas grating.

Maximum laser energy flux we can use in the ozone-mixed gas grating is aslarge as 1.6 kJ/cm2 and 2.7 × 1011W/cm2. Additional wave front distortion mayoccur in such high energy flux condition. To learn about this effect, we measuredthe wave front for a case in which the diffracted 0.53 μm laser and diffracting UVlaser are the same as in previous Fig. 8 experiment, and another high-intensity testlaser is used for demonstration of the high-energy flux beam. A 6 ns pulse of 0.53μm wavelength having the pulse energy from 25mJ to 150 mJ is focused onto a100-μm spot with a 400-mm focal length lens. This intense test beam enters intothe ozone-mixed gas grating almost along the diffracted beam and with the sametiming of the diffracted beams. The energy fluence of this test beam area is from0.4 kJ/cm2 to 2 kJ/cm2 so that we can check both lower and higher intensityconditions of the nominal laser damage threshold (Fig. 4). As shown in Fig. 9, thereis no distortion of the measured fringes when the test laser intensity is lower thanthe nominal optical damage threshold. However there is severe distortion in thewave front at higher intensity test beam conditions. In conclusion, there is nodistortion of wave front if the laser intensity is lower than the nominal breakdownthreshold.

Nonlinear optical effect in gas. The gas (mainly oxygen gas) refractive index isclose to unity (n− 1 < 10−5). Therefore, nonlinear refractive index is quite smallcompared to normal solid optics. The expected minimum threshold of the non-linear optics is rotational stimulated Raman scattering (r-SRS) for a broad-bandlaser. If we use narrower band laser than 0.1 cm−1, stimulated Brillouin scatteringis also a candidate for nonlinear process to change the diffraction condition of thisozone grating. However, the threshold for these effects is factors of ten higher thanour present operating condition. Therefore, we generally can neglect nonlinearoptical effects in gas medium grating with nanosecond operation.

This situation is drastically changed if the pulse duration goes to thefemtosecond level. In our 30 fs experiments, self-phase modulation is lowestnonlinear effect. That threshold energy density is about 6 J/cm2 for a wavelength ofλ= 800 nm. It is also apparent that if the working gas of ozone-mixed gas was ahigher damage threshold gas, such as noble gas, this threshold will be improved.

Fig. 8 Wave front of the diffracted beam. a 0th order non-diffracted beam and b 1st diffracted beam.

Fig. 9 Distortion of the diffracted beam wavefront above nominal damage threshold intensity. Measured interference fringe of 1st order diffractionbeam with an intense (>1 kJ/cm2) damage test beam located at the center of this fringe area. The spot diameter of the intense damage test laser is100 μm, while that of the measured diffracted beam is 4 × 5mm. There is no observable distortion of the fringe under the damage threshold (energyfluence= 1.6 kJ/cm2) (a), but we clearly see the distortion above the damage threshold (>1.6 kJ/cm2) (b).

Fig. 7 Incident angle dependence of the diffraction efficiency. Blue circlesshow the measured value and the red line shows the calculated one. Theerror bar means the standard deviation.

ARTICLE COMMUNICATIONS PHYSICS | https://doi.org/10.1038/s42005-020-0286-6

6 COMMUNICATIONS PHYSICS | (2020) 3:24 | https://doi.org/10.1038/s42005-020-0286-6 | www.nature.com/commsphys

Data availabilityThe raw experimental data are available from the corresponding authors upon reasonablerequest.

Received: 13 December 2019; Accepted: 3 January 2020;

References1. Hurricane, O. A. et al. Fuel gain exceeding unity in an inertially confined

fusion implosion. Nature 506, 343–348 (2014).2. Tajima, T. & Dawson, J. M. Laser electron accelerator. Phys. Rev. Lett. 43, 267

(1979).3. Wagner, C. & Harned, N. EUV lithography: Lithography gets extreme. Nat.

Photonics 4, 24–26 (2010).4. Park, H.-S. et al. High-adiabat high-foot inertial confinement fusion implosion

experiments on the national ignition facility. Phys. Rev. Lett. 112, 055001(2014).

5. Tolenis, T. et al. Next generation highly resistant mirrors featuring all-silicalayers. Sci. Rep. 7, 10898 (2017).

6. Chambonneau, M. & Lamaignère, L. Multi-wavelength growth of nanosecondlaser-induced surface damage on fused silica gratings. Sci. Rep. 8, 891 (2018).

7. Bonod, N. & Neauport, J. Diffraction gratings: from principles to applicationsin high-intensity lasers. Adv. Opt. Photon. 8, 156–199 (2016).

8. Spaeth, M. L. et al. Optics recycle loop strategy for NIF operations above UVlaser-induced damage threshold. Fusion Sci. Technol. 69, 265–294 (2017).

9. Sheng, Z.-M., Zhang, J. & Umstadter, D. Plasma density gratings induced byintersecting laser pulses in underdense plasmas. Appl. Phys. B 77, 673–680(2003).

10. Suntsov, S., Abdollahpour, D., Papazoglou, D. G. & Tzortzakis, S.Femtosecond laser induced plasma diffraction gratings in air as photonicdevices for high intensity laser applications. Appl. Phys. Lett. 94, 251104(2009).

11. Yang, X. et al. Femtosecond laser pulse energy transfer induced by plasmagrating due to filament interaction in air. Appl. Phys. Lett. 97, 071108 (2010).

12. Liu, Y. et al. Energy exchange between femtosecond laser filaments in air.Phys. Rev. Lett. 105, 055003 (2010).

13. Shi, L. et al. Generation of high-density electrons based on plasma gratinginduced Bragg diffraction in air. Phys. Rev. Lett. 107, 095004 (2011).

14. Thaury, C. et al. Plasma mirrors for ultrahigh-intensity optics. Nat. Phys. 3,424–429 (2007).

15. Pinnick, R. G. et al. Aerosol-induced laser breakdown thresholds: wavelengthdependence. Appl. Opt. 27, 987–996 (1988).

16. Michaelis, M. M. et al. A gas-lens telescope. Nature 353, 547–548 (1991).17. Barden, S. C., Arns, J. A., Colburn, W. S. & Williams, J. B. Volume-phase

holographic gratings and the efficiency of three simple volume-phaseholographic gratings. Publ. Astron. Soc. Pac. 112, 809–820 (2000).

18. Daumont, D., Brion, J., Charbonnier, J. & Malicet, J. Ozone UV spectroscopyI: absorption cross-sections at room temperature. J. Atmos. Chem. 15, 145–155(1992).

19. Brion, J. et al. Absorption spectra measurements for the ozone molecule in the350–830 nm region. J. Atmos. Chem. 30, 291–299 (1998).

20. Green, J. G., Shi, J. & Barker, J. R. Photochemical kinetics of vibrationallyexcited ozone produced in the 248 nm photolysis of O2/O3 mixtures. J. Phys.Chem. A 104, 6218–6226 (2000).

21. Chýlek, Petr et al. Effect of spherical particles on laser-induced breakdown ofgases. Appl. Opt. 26, 760–762 (1987).

22. Volkov, V. A. et al. Dependence of threshold for air breakdown by a focusedlaser beam on the geometry of the focal region. Sov. J. Exp. Theor. Phys. 42,58–61 (1976).

23. Tambay, R. & Thareja, R. K. Laser induced breakdown studies of laboratoryair at 0.266, 0.355, 0.532, and 1.06 μm. J. Appl. Phys. 70, 2890 (1991).

AcknowledgementsWe would like to thank Richard. M. More for deep discussion about hydrodynamics.

Author contributionsBoth Y.M. and H.Y. constructed the experiments, as well as analyzed the data, andcontributed to writing the paper.

Competing interestsThe authors declare no competing interests.

Additional informationSupplementary information is available for this paper at https://doi.org/10.1038/s42005-020-0286-6.

Correspondence and requests for materials should be addressed to H.Y.

Reprints and permission information is available at http://www.nature.com/reprints

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims inpublished maps and institutional affiliations.

Open Access This article is licensed under a Creative CommonsAttribution 4.0 International License, which permits use, sharing,

adaptation, distribution and reproduction in any medium or format, as long as you giveappropriate credit to the original author(s) and the source, provide a link to the CreativeCommons license, and indicate if changes were made. The images or other third partymaterial in this article are included in the article’s Creative Commons license, unlessindicated otherwise in a credit line to the material. If material is not included in thearticle’s Creative Commons license and your intended use is not permitted by statutoryregulation or exceeds the permitted use, you will need to obtain permission directly fromthe copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

© The Author(s) 2020

COMMUNICATIONS PHYSICS | https://doi.org/10.1038/s42005-020-0286-6 ARTICLE

COMMUNICATIONS PHYSICS | (2020) 3:24 | https://doi.org/10.1038/s42005-020-0286-6 |www.nature.com/commsphys 7